Elucidation of Exosome Migration across the Blood-Brain Barrier Model In Vitro

Abstract

The delivery of therapeutics to the central nervous system (CNS) remains a major challenge in part due to the presence of the blood-brain barrier (BBB). Recently, cell-derived vesicles, particularly exosomes, have emerged as an attractive vehicle for targeting drugs to the brain, but whether or how they cross the BBB remains unclear. Here, we investigated the interactions between exosomes and brain microvascular endothelial cells (BMECs) in vitro under conditions that mimic the healthy and inflamed BBB in vivo. Transwell assays revealed that luciferase-carrying exosomes can cross a BMEC monolayer under stroke-like, inflamed conditions (TNF-α activated) but not under normal conditions. Confocal microscopy showed that exosomes are internalized by BMECs through endocytosis, co-localize with endosomes, in effect primarily utilizing the transcellular route of crossing. Together, these results indicate that cell-derived exosomes can cross the BBB model under stroke-like conditions in vitro. This study encourages further development of engineered exosomes as drug delivery vehicles or tracking tools for treating or monitoring neurological diseases.

Keywords: Drug delivery; blood-brain barrier (BBB); endocytosis; exocytosis; exosome; humanized Gaussia luciferase (hGluc); inflammation; stroke; transcytosis.

Fig. 1. Exosome engineering and characterization

(a) Schematic depiction of the isolation protocol for exosomes. (b) Size distribution of native and hGluc-Lact exosomes measured by nanoparticle tracking analysis (NTA). (c) Flow cytometry detection of exosome characterization. Exosomes were incubated with anti-CD63 Dynabeads and immunostained with exosomal surface markers (CD9, CD63, and CD81). Green histogram is the isotype control. Data were quantified and expressed as MFI (mean fluorescence intensity). (d) Western blot analysis of the marker proteins (i) CD63, (ii) CD81, and (iii) CD9 and (iv) Gluc on exosomes, respectively. Exosomes were purified from conditioned medium and characterized using western blot for the presence of typical exosomal markers CD63, CD81, and CD9. Additionally, Gluc detection by immunoblot of purified exosomes showed that hGluc-Lact fusion protein bound to the exosomal membranes.

Fig. 2. Validation of hGluc-labeled exosomes using in vitro bioluminescence Assays

(a) In vitro bioluminescence assay of conditioned medium, ultra-centrifugation supernatant and exosomes. (i) Conditioned medium collected from 293T cells, (ii) supernatant collected after serial steps of ultra-centrifugation and (iii) exosomes purified from 293T cells through ultra-centrifugation were diluted in PBS. CTZ was then added at a final concentration of 25 µM. Gaussia luciferase (hGluc) activity was measured using IVIS Lumina (exposure time 0.5s). (b) hGluc activity was significantly higher in conditioned medium. Error bar: mean ± SEM. ****P < 0.0001. (c) After ultra-centrifugation, hGluc activity was mainly detected in hGluc-Lact exosomes, indicating hGluc was enriched on exosomes. Error bar: mean ± SEM. ****P < 0.0001.

Fig. 3. In vitro model of the BBB using BMEC monolayer indicated that stroke-like conditions increased its permeabilit y

(a) The schematic representation of the in vitro model of BBB. (b) A BMECs monolayer was grown for 24 or 48 hours in a transwell insert, and then treated with TNF-α for 6 hours. Permeability was measured using FITC-dextran. BMECs formed a low permeability barrier after 48 hours, and permeability was increased significantly under TNF-α condition. Values represent as means ± SEM of relative ratio normalized to no cell control, set as 100%. **P < 0.01 and ****P < 0.0001 (c) Activation of BMECs with TNF-α regulates tight junction and adherens junction protein expressions in BMEC monolayer on coverglass. Immunofluorescence of VE-cadherin, ZO-1, and Claudin-5 showed that their expression levels were dramatically down-regulated after TNF-α treatment. DAPI was used for staining nuclei. Scale bar: 20 µm.

Fig. 4. Exosomes can cross BMEC monolayer under stroke-like conditions in a transwell assay

(a) Schematic representation of the in vitro model of the BBB. hGluc-Lact exosomes were added to the luminal chamber of the transwell and incubated with BMECs for various time points. Both luminal and abluminal chambers of conditioned medium were collected for bioluminescence assay. (b) and (c) Exosomes can cross BMECs in stroke-like conditions. (b) Representation of in vitro bioluminescence assay. Conditioned medium from both luminal and abluminal chambers were collected after exosome incubation and CTZ was added at a final concentration of 25 µM. Gaussia luciferase activity was measured immediately thereafter using IVIS Lumina (exposure time 0.5s). (c) Quantitative analysis of in vitro bioluminescence assay of hGluc-Lact exosomes crossing both live and fixed BMECs at different time points. Relative Gluc Activity = (abluminal chamber signal - native exo signal) / (luminal chamber signal -native exo signal) × 100%. Error bar: mean ± SEM. Native vs. TNF-α: n.s., not significant, *P < 0.05 and **P < 0.01. Live BMECs (TNF-α) vs. fixed BMECs (TNF-α) at 24 hours: #P < 0.05.

Fig. 5. Validation of exosome crossing BMEC monolayers

(a) Exosomes can cross the BMECs carrying hGluc in vitro. hGluc-Lact exosomes were labeled with the lipophilic dye PKH67, and were added to the luminal chamber of the transwell. Conditioned medium from abluminal chambers were collected and then incubated with a monolayer of BMEC on coverglass to further confirm the hGluc activity observed from bioluminescence assay was directly from exosomes. (b) Exosomes uptake by BMECs. hGluc conditioned medium of abluminal chamber stained with PKH67 was used as a control. Scale bar: 20 µm. (c) The schematic representation of exosome migration from abluminal chamber to the luminal chamber under native and TNF-α-treated conditions. (d) Quantitative analysis of exosome migration from abluminal to luminal chamber at 6 hours and 18 hours. Relative bioluminescence activity suggested that there was no significant difference between native and TNF-α-treated conditions at 6 hours, whereas the relative bioluminescence activity is significant higher in BMECs treated with TNF-α at 18 hours. Relative Gluc Activity = (luminal chamber signal - native exo signal) / (abluminal chamber signal - native exo signal) × 100%. Error bar: mean ± SEM. n.s., not significant and *P < 0.05.

Fig. 6. Exosome uptake by BMECs

(a) and (b) Confocal microscopy analysis of exosome uptake by BMECs. (a) Representative pictures of exosome uptake under normal and stroke-like conditions at selected time points (1, 6 and 18 hour). Exosomes were labeled with PKH67 (green), BMECs were stained with CellMask (deep red), and DAPI (blue) was used for staining nuclei. Scale bar: 20 µm. (b) Quantitative analysis of fluorescence intensity of the PKH67-labeled exosomes. Briefly, the outline of each cell (n > 30) was drawn referring to the cell membrane labeling. The fluorescence intensity of intracellular exosomes that were specifically associated with the cells was then quantified (see Methods). Error bar: mean ± SEM. **P < 0.01.

Fig. 7. Exosome colocalization with early and late endosomes

(a and b) Confocal microscopy analysis shows colocalization of cholera toxin B (CtxB) with exosomes. (a) Representative pictures of colocalization of CtxB with exosomes under normal and stroke-like conditions at 1 and 3 hours. Endosomes of BMECs (red, native or TNF-α stimulated) were stained with CtxB-biotin conjugated with Alexa Fluor 594-streptavidin for 30 minutes and then washed. PKH67-labeled exosomes (green) were incubated with BMECs for 1 or 3 hours. DAPI was used for staining nuclei. Scale bar: 20 µm. (b) Quantitative analysis of colocalization of CtxB with exosomes. Manders’ Colocalization Coefficient denotes the fraction of endosome membrane co-localized with exosomes. The coefficient tends to “1” if the exosomes are highly colocalized with endosomes. Error bar: mean ± SEM. Native vs. TNF-α: *P < 0.05. Native 1 hour vs. 3 hours or TNF-α 1 hour vs. 3 hours: ^#P < 0.05. (c and d) The colocalization of early (Tfn) and late (CtxB) endosomes and exosomes. (c) Representative pictures of colocalization of (i) CtxB and (ii) Tfn with exosomes under stroke-like conditions at 3 hours. PKH67-labeled exosomes (green) were incubated with TNF-α stimulated BMECs for 3 hours, and then endosomes of BMECs (red) were stained with CtxB-Alexa Fluor 594 or Tfn-Texas Red; DAPI was used for staining nuclei. Scale bar: 20 µm. (d) Quantitative analysis of (c) to measure the association of exosomes with CtxB or Tfn. Error bar: mean ± SEM. n.s., not significant.

Fig. 8. Exosomes were exocytosed by BMECs in a transwell assay

(a) Illustration of the pulse-chase experiment to study exosome exocytosis in vitro. (b) Conditioned medium from BMECs pulsed for 6 hours with hGluc-labeled exosomes and chased with unlabeled exosomes (at indicated time points) was collected from both luminal and abluminal transwell chambers at indicated time points and hGluc activity was measured using IVIS Lumina (CTZ final concentration: 25 µM, exposure time: 0.5s). Error bar: mean ± SEM. Native vs. TNF-α: *P < 0.05 and **P < 0.01. Under TNF-α condition: ^####P < 0.0001, as determined by one-way ANOVA with SNK post-hoc test.

Fig. 9. Potential mechanisms of exosome crossing the blood-brain barrier (BBB)

Schematic representation of the proposed mechanisms on how exosomes cross the BBB, and diagram of the major endocytic pathways as well as their corresponding inhibitors used in this study.

Fig. 10. Effects of endocytosis inhibitors on exosome internalization

(a) BMECs were treated with the indicated endocytosis inhibitors for 30 minutes, and then cell viability was determined by XTT assay. Relative cell viability was normalized to the vehicle control (no inhibitor) set as 1. Error bar: mean ± SEM. *P < 0.05 and **P < 0.01, as determined by one-way ANOVA with SNK post-hoc test. (b) BMECs were pretreated with inhibitors or vehicle (no inhibitor), followed by incubation with Texas Red-transferrin, Alexa Fluor 594-CtxB, or Alexa Fluor 594-dextran. Cellular uptake was measured by fluorescence microscopy and fluorescence intensity was quantified and displayed in the bar graphs. Error bar: mean ± SEM. n.s., not significant, *P < 0.05, * *P < 0.01, ***P < 0.001 and ****P < 0.0001. (c) BMECs were pretreated with indicated inhibitors: amiloride (1mM), CPZ (15 µM), cytochalasin D (20 µM), filipin III (5 µM), MβCD (5 mM), nystatin (5 µM) for 30 minutes at 37°C. Exosomes labeled with PKH67 were incubated with BMECs at 37°C for 1 hour, and their uptake was imaged with confocal microscopy and quantified as described in Methods. Error bar: mean ± SEM. Native vs. TNF-α: n.s., not significant, **P < 0.01 and ****P < 0.0001. Conditions compared were: native (unstimulated BMECs) in the absence of inhibitors vs. in the presence of inhibitors, and native BMECs vs TNF-α-stimulated BMECs in the absence or presence of inhibitors: *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001, compared to native no inhibitor or TNF-α no inhibitor conditions, respectively.

Fig. 11. Inhibition of endocytosis decreases exosome crossing the in vitro BBB

BMECs were pretreated with indicated inhibitors: amiloride (1mM), CPZ (15 µM), cytochalasin D (20 µM), filipin III (5 µM), MβCD (5 mM), nystatin (5 µM) for 30 minutes at 37°C, respectively. hGluc-Lact exosomes were subsequently added to the luminal chamber of each transwell and incubated with BMECs for various time points (6 and 18 hours). Cells treated with vehicles (no inhibitor) alone were used as a negative control. To study the temperature effect on endocytosis, BMECs containing exosomes were incubated at either 37°C or 4°C for (a) 6 and (b) 18 hours, and then conditioned medium from both luminal and abluminal chambers were collected and Gaussia luciferase activity was measured immediately after addition of its substrate CTZ (IVIS Lumina, exposure time: 0.5s). Relative Gluc Activity = (abluminal chamber signal -native exo signal) / (luminal chamber signal - native exo signal) × 100%. Error bar: mean ± SEM. Native vs. TNF-α: n.s., not significant, *P < 0.05 and **P < 0.01. #P < 0.05 and ##P < 0.01, compared to native no inhibitor or TNF-α no inhibitor conditions, respectively.